Measure % of PDFs with attachments and % of PDFs that are linearized.

Bug: chromium:768986
Change-Id: I255fb808efd17a48d9e3760f2d3ed3317196a2bc
Reviewed-on: https://chromium-review.googlesource.com/716306
Commit-Queue: Henrique Nakashima <hnakashima@chromium.org>
Reviewed-by: Robert Kaplow <rkaplow@chromium.org>
Reviewed-by: Jochen Eisinger <jochen@chromium.org>
Cr-Commit-Position: refs/heads/master@{#509861}

pdf/out_of_process_instance.cc

@@ -1497,6 +1497,10 @@ void OutOfProcessInstance::DocumentLoadComplete(
   pp::PDF::SetContentRestriction(this, content_restrictions);
   HistogramCustomCounts("PDF.PageCount", document_features.page_count, 1,
                         1000000, 50);
+  HistogramEnumeration("PDF.HasAttachment",
+                       document_features.has_attachments ? 1 : 0, 2);
+  HistogramEnumeration("PDF.IsLinearized",
+                       document_features.is_linearized ? 1 : 0, 2);
 }
 
 void OutOfProcessInstance::RotateClockwise() {

pdf/pdf_engine.h

@@ -61,6 +61,12 @@ class PDFEngine {
   struct DocumentFeatures {
     // Number of pages in document.
     size_t page_count = 0;
+    // Whether any files are attached to document (see "File Attachment
+    // Annotations" on page 637 of PDF Reference 1.7).
+    bool has_attachments = false;
+    // Whether the document is linearized (see Appendix F "Linearized PDF" of
+    // PDF Reference 1.7).
+    bool is_linearized = false;
   };
 
   // The interface that's provided to the rendering engine.

pdf/pdfium/pdfium_engine.cc

@@ -51,6 +51,7 @@
 #include "printing/pdf_transform.h"
 #include "printing/units.h"
 #include "third_party/pdfium/public/fpdf_annot.h"
+#include "third_party/pdfium/public/fpdf_attachment.h"
 #include "third_party/pdfium/public/fpdf_edit.h"
 #include "third_party/pdfium/public/fpdf_ext.h"
 #include "third_party/pdfium/public/fpdf_flatten.h"
@@ -1281,6 +1282,9 @@ void PDFiumEngine::FinishLoadingDocument() {
   if (doc_) {
     DocumentFeatures document_features;
     document_features.page_count = pages_.size();
+    document_features.has_attachments = (FPDFDoc_GetAttachmentCount(doc_) > 0);
+    document_features.is_linearized =
+        (FPDFAvail_IsLinearized(fpdf_availability_) == PDF_LINEARIZED);
     client_->DocumentLoadComplete(document_features);
   }
 }

tools/metrics/histograms/histograms.xml

@@ -58168,11 +58168,27 @@ http://cs/file:chrome/histograms.xml - but prefer this file for new entries.
   </summary>
 </histogram>
 
+<histogram name="PDF.HasAttachment" enum="Boolean">
+  <owner>hnakashima@chromium.org</owner>
+  <summary>
+    Measures if PDFs opened in the PDF viewer have attachments. This is logged
+    whenever a document is opened in the PDF viewer.
+  </summary>
+</histogram>
+
 <histogram name="PDF.IsFontSubstituted" enum="Boolean">
   <owner>npm@chromium.org</owner>
   <summary>
-    Tracks documents opened in the PDF viewer where substitute fonts need to be
+    Measures if PDFs opened in the PDF viewer require fonts to be substituted.
-    used.
+    This is logged whenever a document is opened in the PDF viewer.
+  </summary>
+</histogram>
+
+<histogram name="PDF.IsLinearized" enum="Boolean">
+  <owner>hnakashima@chromium.org</owner>
+  <summary>
+    Measures if PDFs opened in the PDF viewer are Linearized PDFs. This is
+    logged whenever a document is opened in the PDF viewer.
   </summary>
 </histogram>